Scientists from VSLTU named after G. F. Morozov have prepared software and methodological solutions that allow for pre-determining the most fire-vulnerable forest areas, forecasting fire spread, and assessing carbon losses.
The system integrates Earth remote sensing data, drone imagery, and artificial intelligence. Algorithms use AI to determine the condition of tree stands and the viability of plantations. A separate model using UAVs and machine learning assesses the components of forest combustible materials under the canopy — fallen needles, branches, deadwood, and shrub vegetation. Based on this, a methodology for automatic determination of the natural fire danger class for each forest stand has been created.
A separate development direction is related to assessing the consequences of fires. Scientists have created a methodology for calculating the carbon accumulated in the forest ecosystem and proposed combustion coefficients for various components of terrestrial phytomass during fires of different types and intensities. This allows for assessing not only the risk of fire but also the potential carbon losses in case of ignition.
The relevance of such a system is growing: in the European part of Russia, the fire-hazardous period has increased by two to four weeks in recent years. According to the statistics provided by the developers, annually in Russia there are from 9 thousand to 35 thousand forest fires, covering from 500 thousand to several million hectares. According to Rosleskhoz data, the average damage from forest fires is about 20 billion rubles per year.
The developed tools can be integrated into existing forest fire monitoring systems. The next step will be the development of remote assessment of forest conditions and terrestrial phytomass using artificial intelligence, so that forest services can make decisions even before a fire starts.